Learning from Uninterpreted Experience in the SSH

Benjamin Kuipers, Patrick Beeson, Joseph Modayil, and Jefferson Provost

We are working within the framework of the Spatial Semantic Hierarchy (SSH) (Kuipers 2000), which is a lattice of related representations for large-scale space. The SSH is conducive to research on learning from uninterpreted experience because it separates the interface representations -- the representations for local control laws and local metrical models -- from the symbolic representations built on them. We use two robot platforms in this research, both with multiple, partially redundant, sensory modalities. Lassie is an RWI Magellan with laser range-finder, 16 sonar sensors, 16 IR range sensors, 16 bump sensors and odometry. Vulcan is a custom-built robotic wheelchair with two laser rangefinders, binocular cameras on pan-tilt heads, 7 sonars, 12 IR proximity sensors, and odometry.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.